Method for recording storage I/O (input/output) requirement and verifying pressure simulation

Abstract

The invention provides a method for recording a storage I / O (input / output) requirement and verifying pressure simulation. The method comprises the following steps that: firstly, requirements (of different client terminals) on storing I / O resources are virtually recorded in a VDI (virtual device interface) environment, a recording file is transferred to a pressure verifying terminal through a server, a total recording file is replayed to simulate the real pressure, and a replay file is generated; the server analyzes the total recording file and the total replay file, and provides a conclusion that whether the resource configuration states of virtual machines on the pressure verifying terminal meet the requirements on the requirements on the resources of client terminals, and the basis for designing an I / O storage scheme is provided, so that the resource surplus or resource insufficience is avoided, and the applicability of the designed storage scheme is ensured; and meanwhile, the requirement recording and the multithreading pressure replay simulation are automatic, so that the work quantity of requirement acquirement and scheme design is reduced.

Description

【Technical field】 [0001] The invention relates to a method for storage I / O demand recording and pressure simulation verification in a VDI environment. 【Background technique】 [0002] In the desktop virtualization (VDI) solution, the desktop operating system and user data of all users are stored on the storage array in the data center (independent storage or the local disk array of the data center server), and the I / O performance of the storage array is It is a key factor that directly affects the performance of the desktop operating system. During the VDI solution design process, factors such as cost, I / O performance, and capacity of the storage array need to be considered comprehensively in order to strike a balance between these factors. At present, most solution vendors adopt the following methods in the solution design process for the collection and storage solution design of storage array I / O requirements: [0003] 1. There are two main methods for collecting storage a...

AI Technical Summary

Problems solved by technology

[0004] (1) Use the performance counters of the Windows system on a single terminal to collect the storage I / O performance requirements of desktop users. This method requires administrators to manually set performance counters on multiple terminals to collect performance demand data, and the operation is complicated. And the data collected through Windows performance counters can only be viewed, and cannot be directly used for stress verification tests;

[0005] (2), use dedicated software (this software is based on the WMI interface of Windows system) to collect and record the storage I / O performance requirements of desktop users, this method can realize the unified collection of multiple desktop users in the central console I / O performance requires data, but these data can only be simply used to view the I.O usage of a single machine, and cannot be used for data playback

[0007] (1) The design method based on empirical data, this method is based on past experience data for scheme design, and most of the empirical data is based on the data of the traditional PC application mode, and the application characteristics of different customers are different. Therefore, the design method based on empirical data cannot truly reflect the I / O performance requirements of users in the desktop virtualization solution, which makes the designed solution have problems such as excess resources and insufficient resources, which affects the overall solution cost and overall performance

[0008] (2), based on the average value of the data collected by the demand collection method described in the first point in the background technology, this method has some shortcomings, such as: the average value cannot reflect the peak-valley law of user demand;

However, the storage processors (SP) of different types of storage arrays have different processing capabilities and different caches, so the temporary caching and merging capabilities for user I / O operations are also different. The number of hard disks required on the array will also have a large difference, so this design method cannot accurately calculate the storage configuration requirements, and can only be designed based on the collected data based on experience, so there is also a surplus of resources , insufficient resources, etc.

[0010] The publication number is 102393829A, and the publication date is 2012-03-28. The invention patent "a disk I / O test system for embedded systems". I / O acquisition, its I / O acquisition is directly facing the physical hardware layer, and the cloud computing virtualization mode adds a virtual operating system layer on the hardware level, so in the cloud computing virtualization mode, conventional I / O The O acquisition tool cannot directly acquire the I / O of the underlying physical hardware layer in the virtual environment, because the process of calling the physical hardware layer by the virtual operating system layer will consume I / O resources, and the lost I / O resources are passed through conventional I / O acquisition tools is not available

Images